smithery/minitap-ai

mobile-use-setup

Set up a local-first Minitap mobile-use SDK project for iOS or Android, including local devices, BrowserStack, Minitap cloud devices, and supported LLM providers.

Installation

$ npx skills add smithery/minitap-ai --skill mobile-use-setup

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Skill metadata

Parsed from SKILL.md frontmatter.

Declared agents claude-code

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 2,378 B
  • docs SUMMARY.md 223 B

History

  1. First recorded snapshot · 0 installs

SKILL.md

Mobile-Use SDK Setup

Set up mobile-use so the agent runs locally against a selected device.

Gather Requirements

Ask only for information not already provided:

  1. Target: iOS, Android, or both.
  2. Device path: local physical device/emulator, BrowserStack, or a Minitap cloud device.
  3. LLM provider: Minitap API, OpenAI, Anthropic, Google, OpenRouter, a local model, or another supported provider.

Check Prerequisites

python3 --version  # Requires 3.12+
which uv

# Android local
which adb
adb devices

# iOS simulator
which idb_companion
xcrun simctl list devices

# iOS physical
which idevice_id
which appium
appium driver list

For missing dependencies, use the device-specific reference files in this skill.

Create The Project

uv init <project-name>
cd <project-name>
uv add minitap-mobile-use python-dotenv
cp llm-config.override.template.jsonc llm-config.override.jsonc

Configure the selected provider in llm-config.override.jsonc and put only the required provider keys in .env. Add .env to .gitignore.

Create The Starter

Generate task code that runs the agent locally:

import asyncio

from dotenv import load_dotenv

from minitap.mobile_use.sdk import Agent
from minitap.mobile_use.sdk.builders import Builders
from minitap.mobile_use.sdk.types import AgentProfile

load_dotenv()


async def main() -> None:
    profile = AgentProfile(name="default", from_file="llm-config.override.jsonc")
    config = Builders.AgentConfig.with_default_profile(profile).build()
    agent = Agent(config=config)

    try:
        await agent.init()
        result = await agent.run_task(goal="Your automation goal", name="first-task")
        print(result)
    finally:
        await agent.clean()


if __name__ == "__main__":
    asyncio.run(main())

When selected, extend config with the BrowserStack or forclouddevice(...) builder API.

Verify

Verify the selected device is visible, then run:

uv run python -c "from minitap.mobile_use.sdk import Agent; print('SDK OK')"
uv run python main.py

Diagnose setup failures from actual command output. Do not silently switch device paths.